This article reviews recent advances in intelligent robotic ultrasound (US) imaging systems. It begins by presenting commonly used robotic mechanisms and control techniques in robotic US imaging, along with their clinical applications. It then focuses on the deployment of machine learning techniques in the development of robotic sonographers, emphasizing key developments aimed at enhancing the intelligence of these systems. The methods for achieving autonomous action reasoning are categorized into two approaches: those relying on implicit environmental data interpretation and those using explicit interpretation. The article also discusses practical challenges, including the scarcity of medical data, the need for a deeper understanding of the physical aspects involved, and effective data representation approaches. It concludes by highlighting open problems in the field and analyzing different possible perspectives on how the community could move forward in this research area. The article discusses the hardware and control algorithms in robotic ultrasound systems (RUSS), focusing on mechanism design and control methods. It highlights studies using machine learning technologies to improve the understanding of complex natural and physiological scenes or to directly reason proper actions based on implicit scene comprehension. It also examines potential solutions to tackle practical challenges, including the shortage of medical data, the necessity for considering physical aspects, and the implementation of effective data representation methods. The article categorizes machine learning approaches in RUSS into modular and direct approaches for action reasoning. Modular approaches involve integrating the outcomes of deep learning models into rule-based control laws, while direct approaches involve learning from demonstrations or interactions. The article discusses the application of segmentation and registration techniques in robotic ultrasound, highlighting their importance in improving the accuracy and efficiency of US imaging. It also discusses the use of physics-inspired machine learning for US, emphasizing the importance of integrating physiological knowledge and US formation principles into the design of deep learning networks. The article addresses the challenges of data scarcity in robotic ultrasound systems and discusses methods for generating synthetic data, such as ultrasound simulation and data augmentation. It also discusses the importance of physics-inspired machine learning for US, emphasizing the need to integrate physiological knowledge and US formation principles into the design of deep learning networks. The article concludes by discussing open challenges and future perspectives in the field of robotic ultrasound, including ethical and regulatory issues and emerging ultrasound imaging systems.This article reviews recent advances in intelligent robotic ultrasound (US) imaging systems. It begins by presenting commonly used robotic mechanisms and control techniques in robotic US imaging, along with their clinical applications. It then focuses on the deployment of machine learning techniques in the development of robotic sonographers, emphasizing key developments aimed at enhancing the intelligence of these systems. The methods for achieving autonomous action reasoning are categorized into two approaches: those relying on implicit environmental data interpretation and those using explicit interpretation. The article also discusses practical challenges, including the scarcity of medical data, the need for a deeper understanding of the physical aspects involved, and effective data representation approaches. It concludes by highlighting open problems in the field and analyzing different possible perspectives on how the community could move forward in this research area. The article discusses the hardware and control algorithms in robotic ultrasound systems (RUSS), focusing on mechanism design and control methods. It highlights studies using machine learning technologies to improve the understanding of complex natural and physiological scenes or to directly reason proper actions based on implicit scene comprehension. It also examines potential solutions to tackle practical challenges, including the shortage of medical data, the necessity for considering physical aspects, and the implementation of effective data representation methods. The article categorizes machine learning approaches in RUSS into modular and direct approaches for action reasoning. Modular approaches involve integrating the outcomes of deep learning models into rule-based control laws, while direct approaches involve learning from demonstrations or interactions. The article discusses the application of segmentation and registration techniques in robotic ultrasound, highlighting their importance in improving the accuracy and efficiency of US imaging. It also discusses the use of physics-inspired machine learning for US, emphasizing the importance of integrating physiological knowledge and US formation principles into the design of deep learning networks. The article addresses the challenges of data scarcity in robotic ultrasound systems and discusses methods for generating synthetic data, such as ultrasound simulation and data augmentation. It also discusses the importance of physics-inspired machine learning for US, emphasizing the need to integrate physiological knowledge and US formation principles into the design of deep learning networks. The article concludes by discussing open challenges and future perspectives in the field of robotic ultrasound, including ethical and regulatory issues and emerging ultrasound imaging systems.